African Crop Science Journal African Crop Science Society ISSN: 1021-9730 EISSN: 2072-6589 Vol. 4, Num. 2, 1996, pp. 185-196

Relation of some chemical properties to soil erodibility of some south western Kenyan soils

D. W. Macharia, W. Moll^1 and P. A. Kamau*

Egerton University, Agronomy Department, P.O. Box 536, Njoro, Kenya
^1 Institut fuer Bodenkunde und Bodenerhaltung, Wiessenstr. 3-5, 35390 Giessen, Germany

* Corresponding author

(Received 27 June, 1994; accepted 2 May, 1995)

Code Number: CS96056
Sizes of Files:
    Text:
    Graphics: Line drawings (gif) -  12.8K
              Tables (gif) - 63.2K  

ABSTRACT

A laboratory study to investigate the effect of soil cations and dithionite iron (Fe[d]) content on soil erodibility (K) was conducted using simulated rainfall on ten selected Kenyan soils. Rainfall with a simulated intensity of 6 mm min-1 and duration of 3 minutes was applied to soil which had been passed through an 8-mm sieve and packed at a density of 2.5 g cm-3 in trays of dimension 25 25 cm. Soil loss including both splash and wash erosion was determined under three moisture regimes (dry, field capacity (F.C.) and moist/wet). The relevant chemical properties of the soils were analysed and these were correlated with the soil loss. The soils had Fe[d] ranging from 0.2 to 1.5%, manganese (Mn[d]) from 0.31 to 0.80%, silicon (Si[d]) from 0.16 to 0.45% and sodium adsorption ratio (SAR) from 0.02 to 0.33. The total iron (Fe[t]) and aluminium (Alt) contents ranged from 5.9 to 12.6% and from 10 to 23.8%, respectively. The average soil loss varied from 1.8 to 47.4 t ha^-1 and the determined K-values ranged from 0.035 to 0.914. The nomograph K-values overestimated the erodibility of the soils high in Fe[d] and Na, but multiple linear regression analysis confirmed that Fe[d] and Na are important parameters in predicting soil erodibility. About 78% of the variations in soil loss was accounted for by CEC, Exch. Ca, Exch. K, EC and soil organic matter (SOM), whereas 77% of the variations was explained by Ca[t], Fe[d], Fe[t] and Na[t] contents of the soils.

Key Words: Ca[t]ion content, dithionite iron, rainfall simulation

RESUME

Une experience de laboratoire a ete effectuee en utilisant de la precipitation simulee sur dix sols selectionnes au Kenya pour analyser l'effet des cations de sols et de dithionite de fer (Fe[d]) sur l'erosion de terre. Une precipitation avec intensite simulee de 6 mm min-1 et duree de 3 minutes etait donnee e des sols qui ont ete passes au tamis de 8 mm puis presses e une densite de 2,5 g cm^3 dans des plateux de 25-25 cm. La perte de sols par eclaboussement ainsi que par erosion etait determinee pour trois regimes d'humidite (sec, capacite de champ (F.C.) et mouille/humide). Les cracaterisques chimiques utiles des sols ont ete analysees et correlees avec les pertes de sols. Le contenu des sols en cations se rangeait de 0,2 e 1,5% pour le fer (Fe[d]), de 0,31 e 0,80% pour le manganse (Mn[d]), de 0,16 e 0,45% pour le silicium (Si[d]) et le rapport d'absorption de sodium (SAR) de 0,02e 0,33. La teneur totale en Fer (Fe[t]) et aluminium (Alt) se rangeait de 5,9 e 12,6% et de 10 e 23,8%, respectivement. La perte moyenne de terre variait de 1,8 e 47,7 t ha^-1 et la valeur de K de 0.035 e 0.914. Les valeurs nomographiques de K surestimaient l'erosion des sols avec beaucoup de Fe[d] et Na, mais une regression multiple confirme que le Fe[d] et Na sont des parametres importants pour predire l'erosion du sol. Environ 78% de la variabilite en perte de sols est expliquee par le CEC, l'echange de Ca, l'echange de K, EC et la matire organique du sol (SOM), tandis que 77% de la variabilite est expliquee par la teneur de Ca[t], Fe[d], Fe[t], et Na[t] dans les sols.

Mots Cles: Teneur en cation, dithionite de fer, simulation de la precipitation

INTRODUCTION

The soil erodibility constant (K-value) depends on many soil properties ( e.g. texture, chemistry, electrochemical bonds, soil organic matter (SOM), clay mineralogy, permeability and structure (Geng and Coote, 1991; Shafiq et al., 1994). The soils which are high in silt and low in clay and SOM are the most erodible (Ekwue, 1991; Fullen, 1991). Soil erosion losses depend on dispersion and surface sealing of the sodic soils (Miller and Scrifres, 1988; Ela et al., 1992). Highly weathered soils, with very low levels of soluble salts, tend to promote the flocculation of the clay and thus maintain a stable soil structure (Miller et al., 1991). Dispersion and soil crusting result in low infiltration rates, causing soil erosion, and these soil properties are likely to be accelerated by additions of Na through fertilizers or wastes (Shainberg and Levy, 1992; Turski et al., 1992).

The relative importance of soil cations and Fe[d] content in determining soil erodibility was tested by use of a rainfall simulator.

Figure 1

The soils used in the study were sampled from areas already covered by published soil survey reports (Sombroek et al., 1980; Wielemaker and Boxem, 1982; Wamicha et al., 1984). These soils developed in different pedogenic environments and cover seven soil types: Regosol (Udipsamment), Andosol (Dytrandept), Planosol (abruptic Tropaqualfs), Greyzem (Argiaquoll and Hapluroll), Nitisols (Paleustult,Paleustalf and Paleudult), Phaeozem (Hapludoll) and Ferralsol (Haplustox) (Table 1). The soils were collected from South Western Kenya, covering regions at Longonot, Narok, Lolgorien, Kisii, Keroka, Sotik, Kericho and Molo (Fig. 1). Most of the sampling areas have been used extensively for agriculture and grazing. In these areas gulley erosion, which often culminates into deep gorges, is found in the higher hills, whereas sheet and rill erosion predominate on cultivated lower hills. The soils from this region are characterised by strong bonding agents. Some of the soils used in this study were red in colour and upon analysis were found to have a high dithionate iron (Fe[d]) content.

Table 1

The crystallized Fe, Al, Mn and Si oxides were extracted by sodium dithionite as the reducing agent, whereas the weakly crystallized and the exchangeable Fe, Al, Mn and Si were extracted by use of sodium oxalate solution. Sodium pyrophosphate was used to extract organic bonded Fe and Al from the allophane complexes. Clay mineralogy was determined by X-ray diffractometry (Van Reeuwijk, 1986). Aggregate stabilty was determined by mean weighted aggregate diameter (MWD) of the soil aggregates during wet sieving (Hartge and Horn, 1989). Soil texture was determined by the pipette method (Van Reeuwijk, 1986).

Air-dried soils were sieved through 8-mm and packed into trays of size 25 x 25 cm at 2.5 g cm^3 using slight pressure. An average soil layer of 6 cm depth was packed above a 10 cm medium sand layer. Between sand and soil layers was sandwiched a cheese cloth. The sand and the perforations at the bottom of the tray provided drainage for the sample and prevented the downward wash of particles. The soils were placed under an easily hand rotated rainfall simulator (Kamphorst, 1987). The slope angle was held constant at 200 and the rainfall intensity at 6 mm min^-1. The samples were subjected to 3 minutes rainfall and left to drain for 60 minutes. A further 3 minutes of rainfall was followed by a 15 minutes rain-free interval and then a final 3 minutes rainfall period. Each soil was replicated nine times and a completely randomised design was adopted. After each rainfall application, all the material splashed or washed off was collected, dried and weighed. The surface soil was sampled to verify moisture content prior to each rainfall simulation event and the result compared with predetermined pressure plate values. The results were reported as total soil loss under three moisture conditions namely, dry, FC and moist/wet. The K-value was computed from the Universal Soil Loss Equation (USLE) and was also estimated by use of Wischmeier-nomograph (De Meester and Jungerius, 1978; Wischmeier and Smith, 1978). The infiltration rate was computed from the difference between supplied rainfall and runoff (Kamphorst, 1987).

RESULTS AND DISCUSSION

The soils used in the study had textures ranging from loam/loamy sand to clay/clay loam. Only three soils (Longonot, Molo and Sotik) had SOM contents within the 0 - 4% range of the Wischmeier nomograph (Table 1). All the other soils had more than 4% SOM and so their K-values would be poorly predicted by the nomograph. The studied soils had a Fe[d] range from 0.2 to 1.5%, Mn[d] from 0.31 to 0.80%, Si[d] from 0.16 to 0.45% and SAR from 0.02 to 0.33 (Table 2). The results of some of the most important soil characteristics (Tables 1, 2 and 3) were pH (H2O): medium to neutral (5.3 - 7.0); EC: very low to low (0.04 - 0.10 mmhos/cm); ESP: very low to low (0.2 to 4.4); CEC: low to medium (8 - 30.2 me/100 g soil); BS: medium to high (39 - 93%), CaCO3: very low to low (0.01 - 0.21%); SOM: low to very high (1.6 - 9.3%), Fe[t]: medium to high (5.9 - 12.6%) and Al1 medium to high levels (10 - 23.8%). The clay minerals composition was mainly kaolinite and some illite for most of the soils except Lolgorien 1 and 2 soils which had mainly smectite and illite. The above parameters were correlated with soil loss and used in multiple regression in order to determine their effect on soil erodibility.

Figure 2

Table 2

Table 3

Table 4

Table 5

In most of the cases, the soil loss and determined K-values were higher for the second (wet 1: FC) than for the first (dry) storm (Table 4). Most of the soils in the studied area had a low to moderate erosion susceptibility. The Nitisols (Kericho and Kisii except Molo) and Phaeozem (Keroka) had the lowest K-values because of their good aggregate stability and slightly higher infiltration rates. Molo (Nitisol) had a poor aggregate stability and moderate infiltration rate (Table 4). The Ferralsol (Sotik), Greyzems (Lolgorien 1 and 2) and Andosol (Narok 1) soils had low to moderate erodibility, whereas the Regosol (Longonot) and Planosol (Narok 2) soils had higher K-values due to their more restricted drainage.

For the soils relatively rich in iron (Kericho, Kisii, Keroka, Sotik and Molo), and total sodium (Narok 1 and 2, Lolgorien 1 and 2 and Longonot), the nomograph did not predict well the relative erodibilities (Tables 2 and 3). The nomograph K-values (Fig.2) overestimated the erodibility of soils high in Na (except for Longonot - a soil developed from volcanic ash) and Fe[d]. It was also observed that the soils which were highly erodible (Regosol: Longonot; Planosol: Narok 2; Haplic Greyzem: Lolgorien 1; humic Nitisol: Molo) had either a low porosity or a high SOM (Andosol: Narok 1) combined with relatively high sand or silt contents (45 - 51%). It was found that Regosol was the most erodible and the Nitisols (except Molo) were the least erodible soils (Fig.2 and Table 4). Molo soil had relatively lower Fe[d] and Ald (Table 2), infiltration rate (Table 4) and higher Na[t] than other two Nitisols, indicating that Molo was less stable (a fact shown by aggregate stability values).

Infitration rates and Fe[d] content of the studied soils gave similar trend to that shown by K-values, but was found to be inversely proportional to that of Na[t]-content (Tables 4, 2 and 3). These findings emphasise the importance of infiltration rate, Fe[d] and Na in predicting soil erodibility. The K-value increased with Na content but decreased with infiltration rate and Fe[d] content of the studied soils.

Based on the soil properties, results of the simulation study could be interpreted as follows: a low initial water content and a loose surface soil structure led to very high initial infiltration rates. Longonot, Narok 2, Molo and Lolgorien 2 soils generated run-off and soil loss during all or at least the second and third simulation phases: amounts of runoff and soil loss were much higher than in the second group (Sotik, Narok 1 and Lolgorien 1 soils). The higher erodibilities of group 1 soils was characterized by lower permeability for water, a lower content of SOM and a lower aggregate stability in the surface soil (Tables 1 and 4). The second group had intermediate values of these parameters. The third group (Kericho, Kisii and Keroka) of soils absorbed most of the applied rainfall such that runoff developed late and soil loss was low.

The pH (0.2m CaCl2) showed a low but significant correlation with soil loss (r = 0.27***) (Table 5). This result agrees with that reported by Trott et al. (1983) which indicated that soils with high pH are more susceptible to erosion as was the case with Narok 1 soil (Andosol) in this study.

It was observed that the Fe[t] content of the studied soils exceeded the Alt content (Table 3). The Regosol (Longonot) and Planosol (Narok 2) soils had the lowest levels of sesquioxide. All the other soils had the highest sesquioxide, although the soils with extremely high levels of SOM also tended to have high Fe and Al contents.

The dithionite-extractable Al203d and Fe203d were both negatively correlated with soil loss (r = 0.35*** and r = 0.40***, respectively) (Table 5). These results support earlier findings (Despande et al., 1968; Romkens et al., 1977; Nizeyimana and Olson, 1988) that, as sesquioxide levels in soils increased, there was an enhanced aggregate stability and this presumably improved soil structure and so reduced soil erosion.

In South Western Kenya there are many soils with large contents of Fe[d] and others with high Na contents (Macharia, 1992). Iron and sodium were found to be closely correlated with soil erodibility under simulated rainfall. A strong inverse relationship was found between Fe[d] and erodibility. These two chemical properties are important in aggregate stability and they should be useful additions to prediction of soil K-values. These findings support those reported in literature (Singer et al., 1980; Shainberg and Levy, 1992).

Scanning through the soil data, several patterns are observable. The Kericho, Kisii (both Nitisols) and Keroka (Phaeozem) soils all have low average soil losses and low K-values. There is some question as to the K-value of the Narok 1 soil (Andosol) because of its very high SOM content (8.1%). The Molo (Nitisol), Narok 2 (Planosol) and Lolgorien 2 (Greyzem) soils are intermediate between the lowest erodible group and the next group with moderate erodibilities, and the K-values for these soils were in the range of 0.21 to 0.30. Three soils with moderate erodibility levels were Narok 1 (Andosol), Lolgorien 1 (Greyzem) and Sotik (Ferralsol) soils (Singer et al., 1980; Nizeyimana and Olson, 1988).

In two cases (Sotik and Lolgorien 2) K-values were equal to or lower than those of soils with much lower soil losses and in two cases (Narok 1 and Keroka) the K-values were equal to or nearly equal to those of soils with much higher soil losses. Kericho and Keroka soils had almost identical soil loss levels but a considerable difference in K-values. The Longonot and Narok 2 soils had the highest soil loss and were statistically different. Their K-values follow this pattern too. Molo had a K-value equal to Lolgorien 1 but had a soil loss level about two times as high. The same case was observed with Lolgorien 2 and Narok 1 (almost double soil loss level but same K-values). This pattern was also shown by Sotik and Kisii soils (double). Another interesting finding was that of K-value of Narok 1 which was thirty times as high but its soil loss level is about double that of Kericho. Molo, Lolgorien 2, Lolgorien 1 and Narok 1 soils all have almost equal K-values but the soil losses of Molo and Lolgorien 2 are almost two times that of Narok 1, indicating that either the Narok 1 K-values is too high or the Molo and Lolgorien 2 K-values are too low (Trott et al., 1983; Macharia, 1992).

It is apparent from the relatively high SAR of the Lolgorien 1, Lolgorien 2 and Longonot soils, and the current knowledge on the effect of Na on soil dispersion, that Na content of these soils must have an effect on their erodibility. Sodium is not considered in the Wischmeier nomograph (Wischmeier and Mannering, 1969) and it would be worthwhile to begin to attempt to quantify its effect especially in the tropical saline soils which are normally widely spread in the study area and elsewhere. In these cases Lolgorien 1 and 2 had relatively higher SAR values and consequently relatively high soil loss values, although the other soils had almost equal K-values. The erodibility of the natural (Na-rich) Greyzem soils (Lolgorien 1 and 2 ) and Longonot is very high. This shows the importance of Na in soil erodibility (Miller and Scrifres, 1988; Turski et al., 1992).

Iron was in general trend found to be inversely related to soil erodibility. As expected,it is apparent that the soils with high Fe[d] were the least erodible although variation in SOM content and texture tended to be obscured by the effect of Fe. Some researchers (Kemper and Koch, 1966) highlighted the importance of Fe , while others reported Al to be more important in influencing soil erodibility (Deshpande et al., 1968; Geng and Coote, 1991). For this study, both Fe and Al follow the same trend as the soil loss and so they seem to be of similar or equal importance in soil erodibility. A better relationship was obtained between erodibility and amorphous (Feox) and with organic iron (Feorg). Several of the studied soils had Fe[d] contents greater than 1% but also had high soil losses. Molo and Sotik soils also had Fe[d] contents exceeding 1% but this level did not apparently reduce their erodibility. Perhaps the 1% Fe[d] was too low to have an influence on soil erodibility. In case of Lolgorien 1 and 2 soils (Greyzems), the SAR was probably the dominant factor in the erodibility.

In spite of the low erodibility of the studied soils, soil loss was still excessive from these soils when kept bare (Longonot and Narok 2). Soil losses between 1.8 and 47.4 metric tons per hectare per year were measured (Table 4). The critical range has been quoted for Kenyan soils as 6.7 ton^-1 ha^-1 year^-1 (Macharia, 1992). These results agree with the contention that the high soil losses in tropical soils may be determined more by factors like erosivity (especially in the humid tropics), topography, crop and soil management than by soil erodibility (Quansah, 1981; Imeson and Kwaad, 1990 ; Ela et al., 1992).

When compared to the values available in the literature, mean weighted K-values presented here for South Western Kenya soils (from a very low of 0.01 to a moderate of 0.30) cover a range equally as wide as that of temperate soils (Nizeyimana and Olson, 1988; Fullen, 1991; Turski et al., 1992). Erodibilities of soils derived from rather similar parent materials (e.g. Nitisols: Kericho, Kisii and Molo ; and Greyzems: Lolgorien 1 and 2) at different locations generally vary inversely with prevailing annual rainfall at those locations. This would indicate that these erodibilities are strongly related to such properties as aggregate stability, clay and sesquioxide contents and soil acidity, all of which increase in magnitude with high intensity of weathering. The Nitisols are of relatively low erodibility, which is not surprising in view of their high structural stability (Geng and Coote, 1991).

When soils were ranked according to mean soil loss (Table 4), it was observed that the soils in the lower half of the ranking (Kericho, Keroka, Kisii and Narok 1) had either higher levels of SOM, were coarse textured, or more red in colour due to relatively higher amounts of sesquioxide than soils in the upper half. The group of red soils (Keroka and Kericho) produced significantly lower soil losses than the comparison group, despite having a similar structure, SOM and lower permeability . The differences appear to originate from differences in iron oxide contents and perhaps clay mineralogy (Tables 1, 2 and 3).

The high erodibility of some of the South Western Kenyan soils, particularly from semi-arid soils (Longonot and Narok 2), can be explained largely by their poor aggregation. This is due to external influences, such as volcanic ash deposition, and also to characteristics such as low SOM content.

The smectite disrupts aggregate fabric by very high water absorption and swelling, facilitating breakdown (Romkens et al., 1977; Nizeyimana and Olson, 1988; Geng and Coote, 1991). This effect may be counteracted by organic cations or accentuated by Na-cations absorbed on clay surfaces. So the effect of smectite is emphasised in the semi-arid soils (in this study Greyzems: Lolgorien 1 and 2 soils) where SOM contents are low and Na frequently dominates the clay complex. Organic material is also an important bonding agent in aggregates apart from its effect on smectite swelling capacity, and so soils deficient in SOM such as Longonot and Narok 2 tend to be very highly erodible (Table 1).

The low but significant correlation between soil loss and del-pH (i.e., changes in pH = pH (CaCl2) - pH (H2O) explains the importance of pH in volcanic ash solutions such as the Longonot (Regosol) soil in this study (Trott et al., 1983). In volcanic ash soils there is the dominance of amphoteric constituents and also they have higher content of silt and sand both of which are highly erodible. Volcanic ash soils exhibit more structural dependency on del-pH than highly weathered soils containing high clay (e.g. Nitisols for this study). A stepwise multiple linear regression analysis between soil loss values and the important soil parameters as independent variables to derive reliable erodibility prediction for the South Western Kenyan soils was conducted. For this study, chemical properties and mineralogical properties were included in the multiple regression analysis, in addition to some of the properties measured by Wischmeier and Mannering (1969).

The following soil-loss (SL) equations were obtained:

1. SL = - 17.728 + 2.708 CEC - 3.792 Exch.Ca - 11.817 Exch.K + 1418.547 EC - 10.483 OM. R^2 = 0.7785 : 78%

2. SL = -90.644 - 0.072 Ca[t] - 83.527 Fe[d] + 24.683 Fe[t] + 0.034 Na[t] R^2 = 0.07735 : 77%

3. SL = 114.078 + 68.024 Al[2]O[3d] - 446.641 SiO[2ox] - 76.333 SiO[2d] + 74.985 MnO[2t] - 168.983 MnO[2ox] - 1.389 SEC - 1045.532 Al[org] R^2 = 0.7830: 78% )SEC = SUm of Exchangeable Ca[t]ions)

4. SL = 36.126 + 0.068 Na[t] - 1.900 Al[ox]/Al[d] + 5298.544 SiO[2ox] - 0.126 Ca[t] R^2 = 0.2248 : 22%

The multiple linear regression analysis results indicate that the four most important variables in predicting erodibility of these ten soils are Ca[t], Fe[t], Fe[d] and Na[t] . These four factors accounted for 77% of variation in the measured mean weighted soil loss. This finding emphasises the role of sesquioxides in soil erodibility. The results also indicate that both Fe and Na content of these soils are important in predicting the erodibility of some S.W. Kenyan soils, and that these need to be examined further along with other factors for tropical soils. The results agree with those of El-Swaify and Dangler (1977) who worked with Hawaiian soils. Highly weathered tropical soils have lower erodibility than the less weathered soils found in temperate climate (Geng and Coote, 1991; Turski et al., 1992). This was found to be the case with Nitisols (Kericho and Kisii) and Phaoezem (Keroka) soils which are highly weathered and showed low erodibility whereas the Regosol (Longonot) and Planosol (Narok 2) soils, which are less weathered and originated from recent volcanic ash, showed high soil erodibility values. This was attributed to differences in water transmission properties and structural characteristics.

Strong multicollinearities existed between SOM and extractable Al and Fe : r = 0.83, permeability and Fe : r = 0.86, and between several of the sesquioxide terms and SOM. All these were significant (P < 0.1%). Soil properties that were most strongly correlated with soil loss (whether dry, F.C (wet 1), moist/wet (wet 2) or mean weighted K-values) were Na and SOM among many others. These observations are in agreement with those of other workers (Imeson and Kwaad, 1990; Turski et al., 1992). The CEC, illite and Ca[t] were less strongly correlated with soil loss. The least correlated parameters were Kaolinite, Fe[d], SiO[2ox] and pH (CaCl2). Those negatively correlated with soil loss were SiO[2ox], SOM, del-pH, clay, CEC and Fe[d]. CEC was found to be a necessary parameter for prediction of wet soil loss values. Del-pH was only selected in K-wet and K-dry when runoff was introduced in the regression.

About 78% of the variation in soil loss could be explained by the following chemical properties : CEC, Exch. Ca and K, EC and SOM contents. The multiple regression performed between soil loss and oxides of Fe, Al, Mn and Si gave Equation 3 and they accounted for 78% of the variations in soil loss. In the same equation, the sum of exchangeable cations was included but without iron term. Iron was strongly inter-correlated with aluminium and this may explain the disappearance of the iron term in the equation. The importance of Al[org] , Al2O[3d], SiO[2ox], Si[O2d], MnO[2t], Fe[d], Fe[t] and MnO[2ox] as better predictors than the other Fe or Al terms has been indicated in Equations 2 and 3. Sesquioxide terms have been found useful in other models of soil erodibility (Deshpande et al., 1968; El-Swaify and Dangler, 1977; Singer et al., 1980).

The best single predictive factors for the relative erodibilities were the percentage SOM (r= -0.60***), clay (r = -0.60***), sodium (r = 0.79***) and Fe[ox] (r = -0.64***), all being significant at P < 0.1%.

CONCLUSIONS

In South Western Kenya, there are many soils with large contents of Fe[d] and others with large Na contents. These two chemical parameters, together with sesquioxides, were found to be closely correlated with soil erodibility under simulated rainfall. These parameters are important in aggregate stability and they should be useful additions to predictions of soil K-values.

ACKNOWLEDGEMENT

The financial support given to the main author by the Ecumenical Scholarship of Germany during this study is gratefully acknowledged. The constructive critisms from Prof. L. King of Justus-Liebig University, Giessen, Germany, Institute of Geography, Faculty of Geological Sciences and Geography is highly appreciated.

REFERENCES

De Meester, T. and Jungerius, P.D. 1978. The relationship between the soil erodibility factor K (Universal Soil Loss Equation), aggregate stability and micromorphological properties of soils in the Hornos area, Southern Spain. Earth Surface Processes and Landforms 3:379-391.

Deshpande, T.I., Greenland, D.J. and Quirk, J.P. 1968. Changes in soil properties associated with the removal of iron and aluminium oxides. Journal of Soil Science 19: 108-122.

Ekwue, E.I. 1991. The effects of soil organic matter content, rainfall duration and aggregate size on soil detachment. Soil Technology 4:197-207.

Ela, S.D., Gupta, S.C. and Rawls, W.J. 1992. Macropores and surface seal interactions affecting water infiltration into soil. Soil Science Society of America Journal 56:714 -721.

El-Swaify, S.A. and Dangler, E.W. 1977. Erodibilities of selected tropical soils in relation to structural and hydrologic parameters. In: Soil Erosion, Prediction and Control. Proceedings of National Conference in Soil Erosion, Indiana, Special Publication 21: 393 pp.

Fullen, M.A. 1991. Soil organic matter and erosion processes on arable loamy sand soils in the West Midlands of England . Soil Technology 4:19-31.

Geng, G.Q. and Coote, D.R. 1991. The residual effect of soil loss on the chemical and physical quality of three soils. Geoderma 48:415-429.

Hartge, K.H. and Horn, R. 1989. Die physikalische Untersuchung von Boden. 2. Auflage. Ferdinand Enke Verlag, Stuttgart. 289pp.

Imeson, A.C. and Kwaad, F.J.P.M. 1990. The response of tilled soils to wetting by rainfall and the dynamic character of soil erodibility. In: Soil Erosion on Agricultural Land. Boardman, J., Foster, I.D.L. and Dearing, J.A. (Eds.), pp. 3-14. Wiley, Chichester.

Kamphorst, A. 1987. A small rainfall simulator for the determination of soil erodibility. Netherlands Journal of Agricultural Science 35:407-415.

Kemper, W.D. and Koch, E.J. 1966. Aggregate stability of soils from Western United States and Canada. Technical Bulletin No.1355 USDA - ARS. U.S. Government Printing Office, Washington DC.

Macharia, D.W. 1992. The Relative Erodibility of Typical South-western Kenyan Soils and the Possibility of Stabilizing Them. Band 10. Giessener Bodenkundliche Abhandlungen. Institute fuer Bodenkunde und Bodenerhaltung der Justus-Liebig-Universitaet Giessen. 177pp.

Miller, W.P. and Scrifres, J. 1988. Effect of sodium nitrate and gypsum on infiltration and erosion of a highly weathered soil. Soil Science 145:304-309.

Miller, W.P., Summer, M.E. and Kim K.H. 1991. Chemical amelioration of surface crusting to reduce runoff and erosion on highly weathered soils. Soil Technology 4:319-327.

Nizeyimana, E. and Olson, K.R. 1988. Chemical, mineralogical and physical property differences between moderately and severely eroded Illinois soils. Soil Science Society of America Journal 52:1740-1748.

Quansah, C. 1981. The effect of soil type, slope, rain intensity and their interactions on splash detachment and transport. Journal of Soil Science 32:215-224.

Romkens, M.J.M., Roth, C.B. and Nelson, D.W. 1977. Erodibility of selected clay subsoils in relation to physical and chemical properties. Soil Science Society of America Journal 41:954-960.

Shafiq, S., Hassan, A. and Ahmad, S. 1994. Soil physical properties as influenced by induced compaction under laboratory and field conditions. Soil Tillage Research 29:13-22.

Shainberg, I. and Levy, G.J. 1992. Physico-chemical effects of salts upon infiltration and water movement in soils. In: Interacting Processes in Soil Science. Wagenet, R.J., Baveye, P. and Stewart, B.A.(Eds.), pp. 37-94. Advances in Soil Science. Lewis Publishers.

Singer, M.J., Blackard, J. and Janitzky, P. 1980. Dithionite iron and soil cation content as factors in soil erodibility. In: Assessment of Soil Erosion. De Boodt, M. and Gabriels, D. (Eds.), pp. 259-268. John Wiley and Sons Inc., New York.

Sombroek, W.G., Braun, N.M.H. and van der Pouw, G.J.M. 1980. Exploratory Soil Survey Report No.1, Kenya Survey Report No.46, Nairobi.

Trott, K.E. and Singer, M.J. 1983. Relative erodibility of 20 California range and forest soils. Soil Science Society of America Journal 47:753-759.

Turski, R., Paluszek, J. and Slowinska-Jurkiewicz, A. 1992. The effect of erosion on the spatial differentiation of the physical properties of orthic Luvisols. International Agrophysics 6:123-136.

Van Reeuwijk, L.P. 1986. Procedures for Soil Analysis. International Soil Reference and Information Centre (ISRIC). Technical Paper 9. Wageningen, Netherlands.

Wamicha, W.N., Kiome, L.M. and Okoth, P.F. 1984. Soils of the Transmara Division. Kenya Soil Survey Report No.56. Nairobi.

Wielemaker, W.G. and Boxem, H.W. 1982. Soils of the Kisii area. Reconnaissance Soil Survey Report No. R4. Kenya Soil Survey Report, Nairobi.

Wischmeier, W.H. and Mannering, J.V. 1969. Relation of soil properties to its erodibility. Soil Science Society of America Proceedings 33:131-137.

Wischmeier, W.H. and Smith, D.D. 1978. Predicting rainfall erosion losses - a guide to conservation planning. USDA Agricultural Handbook No. 537.

Copyright 1996 The African Crop Science Society

The following images related to this document are available:

Line drawing images